Exposure to crime and violence can harm individuals' quality of life and the
economic growth of communities. In light of the rapid development in machine
learning, there is a rise in the need to explore automated solutions to prevent
crimes. With the increasing availability of both fine-grained urban and public
service data, there is a recent surge in fusing such cross-domain information
to facilitate crime prediction. By capturing the information about social
structure, environment, and crime trends, existing machine learning predictive
models have explored the dynamic crime patterns from different views. However,
these approaches mostly convert such multi-source knowledge into implicit and
latent representations (e.g., learned embeddings of districts), making it still
a challenge to investigate the impacts of explicit factors for the occurrences
of crimes behind the scenes. In this paper, we present a Spatial-Temporal
Metapath guided Explainable Crime prediction (STMEC) framework to capture
dynamic patterns of crime behaviours and explicitly characterize how the
environmental and social factors mutually interact to produce the forecasts.
Extensive experiments show the superiority of STMEC compared with other
advanced spatiotemporal models, especially in predicting felonies (e.g.,
robberies and assaults with dangerous weapons).

本文提出了一种空间时间元路径引导的可解释犯罪预测（STMEC）框架，通过捕捉社会结构、环境和犯罪趋势的信息，明确地表征环境和社会因素相互作用产生预测。大量实验表明 STMEC 在预测罪行（如抢劫和使用危险武器的攻击）方面优于其他先进的时空模型。

时空元路径引导的可解释犯罪预测

Spatial-Temporal Meta-path Guided Explainable Crime Prediction

Correlations between anomalous activity patterns can yield pertinent
information about complex social processes: a significant deviation from normal
behavior, exhibited simultaneously by multiple pairs of actors, provides
evidence for some underlying relationship involving those pairs---i.e., a
multilateral relation. We introduce a new nonparametric Bayesian latent
variable model that explicitly captures correlations between anomalous
interaction counts and uses these shared deviations from normal activity
patterns to identify and characterize multilateral relations. We showcase our
model's capabilities using the newly curated Global Database of Events,
Location, and Tone, a dataset that has seen considerable interest in the social
sciences and the popular press, but which has is largely unexplored by the
machine learning community. We provide a detailed analysis of the latent
structure inferred by our model and show that the multilateral relations
correspond to major international events and long-term international
relationships. These findings lead us to recommend our model for any
data-driven analysis of interaction networks where dynamic interactions over
the edges provide evidence for latent social structure.

介绍了一种新的非参数贝叶斯潜变量模型，该模型明确捕捉了异常交互计数之间的相关性，并利用这些共享的偏差从正常活动模式中识别和表征多边关系。通过使用新近策划的全球事件位置与语气数据库，显示了该模型的能力，并在研究结果中指出多边关系对应于重大国际事件和长期国际关系，因此建议在数据驱动的交互网络分析中应用该模型。

从动态配对互动推断多边关系

Inferring Multilateral Relations from Dynamic Pairwise Interactions

We introduce the graphlet decomposition of a weighted network, which encodes
a notion of social information based on social structure. We develop a scalable
inference algorithm, which combines EM with Bron-Kerbosch in a novel fashion,
for estimating the parameters of the model underlying graphlets using one
network sample. We explore some theoretical properties of the graphlet
decomposition, including computational complexity, redundancy and expected
accuracy. We demonstrate graphlets on synthetic and real data. We analyze
messaging patterns on Facebook and criminal associations in the 19th century.

本文介绍了使用图结构对加权网络进行图元分解的方法，并使用具有可扩展性的 EM-Bron-Kerbosch 算法对其进行参数估计，探究了图元分解的计算复杂性、冗余性和预期精度等理论性质，并在 Facebook 的消息模式和 19 世纪的犯罪组织等实际数据上进行了应用和分析。